A method and system for identifying abnormality of transformer oil temperature meter

Abstract

The present invention relates to the technical field of transformer detection, and discloses a method and system for identifying abnormalities of a transformer oil temperature meter. The method constructs an oil temperature prediction sample set through oil temperature data, load data and ambient temperature data of the transformer, trains the oil temperature prediction sample set based on a support vector regression model that performs feature selection and parameter optimization using a genetic algorithm, constructs a transformer oil temperature prediction model, and performs transformer oil temperature prediction. The static state identification result of the transformer oil temperature meter is calculated from three aspects: single point deviation, sequence fluctuation and residual distribution of the transformer oil temperature meter. A sliding time window is also used to continuously monitor the changing trend of the oil temperature residual sequence, and based on this, an abnormal dynamic identification result of the transformer oil temperature meter is calculated. The state of the transformer oil temperature meter is identified by integrating the static state identification result and the abnormal dynamic identification result of the transformer oil temperature meter, thereby improving the accuracy and efficiency of abnormal identification of the transformer oil temperature meter.

Description

Technical Field

[0001] The present invention relates to the technical field of transformer detection, and in particular to a method and system for identifying abnormality of a transformer oil temperature meter. Background Art

[0002] Power transformers are crucial components of the power system. Their safe and stable operation is essential for ensuring the normal operation of the power grid, power supply, and society. In particular, a transformer's load capacity and insulation aging rate are primarily determined by its thermal characteristics, and oil temperature is a key indicator of this.

[0003] The design and manufacturing team installed an oil temperature meter on the transformer and transmitted the oil temperature data to the dispatching master station via the SCADA system, enabling remote, centralized monitoring of the transformer's oil temperature. During daily operation, staff monitor oil temperature fluctuations in real time to understand the transformer's operating status. If an oil temperature exceeding the limit alarm is issued, measures such as load adjustment and professional inspections are implemented to ensure safe and stable operation of the transformer. Therefore, the proper and reliable operation of the oil temperature meter is crucial to ensuring safe and reliable equipment operation.

[0004] Currently, power supply companies periodically perform on-site calibration of transformer oil temperature meters. This process monitors meter status and identifies abnormal meters, ensuring proper functioning. However, periodic calibration makes it difficult to monitor the meter's operating status in real time, creating a "vacuum" in monitoring and hindering the development of an intrinsically safe system for the equipment. Furthermore, the increasing scale of equipment and on-site workloads are increasing, placing greater pressure on frontline teams. Furthermore, current transformer oil temperature monitoring data is transmitted remotely to the dispatching master station via SCADA. However, insufficient data mining is being applied to transformer oil temperature monitoring data, resulting in low accuracy and efficiency in identifying abnormalities in transformer oil temperature meters. Summary of the Invention

[0005] The present invention provides a transformer oil temperature meter abnormality identification method and system, which solves the technical problem of low accuracy and efficiency in transformer oil temperature meter abnormality identification.

[0006] In view of this, a first aspect of the present invention provides a method for identifying an abnormality of a transformer oil temperature meter, comprising the following steps:

[0007] S1. Obtain transformer oil temperature data, load data, and ambient temperature data to construct an oil temperature prediction sample set;

[0008] S2. Training the oil temperature prediction sample set based on a support vector regression model that has been subjected to feature selection and parameter optimization using a genetic algorithm to construct a transformer oil temperature prediction model;

[0009] S3, performing transformer oil temperature prediction based on the transformer oil temperature prediction model, calculating the single point deviation, sequence fluctuation and residual distribution of the transformer oil temperature meter's residual sequence according to the oil temperature prediction result, and obtaining a corresponding static state identification result of the transformer oil temperature meter;

[0010] S4. Monitor the oil temperature prediction results using a sliding time window to obtain a change trend of the transformer oil temperature residual sequence, and calculate an abnormal dynamic identification result of the transformer oil temperature meter based on the change trend of the transformer oil temperature residual sequence;

[0011] S5. Identify the status of the transformer oil temperature meter by integrating the static status identification results and abnormal dynamic identification results of the transformer oil temperature meter.

[0012] Preferably, step S1 specifically includes:

[0013] The transformer oil temperature data, load data and ambient temperature data are obtained through the dispatching EMS system. The current load, historical load, current ambient temperature, historical ambient temperature and historical oil temperature are used as the input of oil temperature prediction, and the oil temperature at future moments is used as the output of oil temperature prediction to construct an oil temperature prediction sample set.

[0014] Preferably, step S2 specifically includes:

[0015] S201, the oil temperature prediction sample set is used as a training sample set (x j ,y j ), j = 1, 2, ..., J, x∈R N , y∈R, where x j is an N-dimensional input vector, y j is the output vector, R N is the feature space of the input vector, J is the number of training samples, and the regression estimation function constructed in the high-dimensional space is:

[0016]

[0017] Where: w is the weight coefficient; b is the threshold; is a nonlinear mapping function;

[0018] S202. Convert the regression estimation function into the objective function and inequality constraints of the quadratic programming problem:

[0019]

[0020] st|y j -(w T x j +b)|≤ε,j=1,2,…,J

[0021] Where T is the transposition symbol and ε is the slack variable;

[0022] S203, introduce slack variable ξ j , transform the inequality constraints of the quadratic programming problem into:

[0023] |y j -(w T x j +b)|≤ε+ξ j

[0024] S204. Introduce the penalty factor C and convert the objective function of the quadratic programming problem into:

[0025]

[0026] S205. For the linear inseparable problem, the radial basis kernel function is used to perform data mapping processing, and the result is:

[0027] K(x,x j )=exp(-γ||xx j || 2 )

[0028] Where γ is the parameter that controls the scale of the kernel function;

[0029] S206 , optimizing the input vector features, the support vector regression parameter C, and the parameter γ of the control kernel function scale based on a genetic algorithm, and performing oil temperature prediction based on the optimized transformer oil temperature prediction model.

[0030] Preferably, the step of optimizing the input vector features, the support vector regression parameter C, and the parameter γ for controlling the kernel function scale based on the genetic algorithm specifically includes:

[0031] Randomly generate binary codes to form the initial population. The chromosomes in the initial population are composed of three parts: input vector features, support vector regression parameters C, and parameter γ that controls the kernel function scale.

[0032] Convert the chromosome from binary code to decimal value using the following formula:

[0033]

[0034] Where p is the decimal value of the parameter corresponding to the chromosome gene sequence; p min 、p max are the minimum and maximum values ​​of the parameters respectively; d is the decimal value of the chromosome gene sequence; l is the length of the chromosome gene sequence;

[0035] The fitness value of the chromosome is calculated by the fitness function of the following formula:

[0036]

[0037] Where, fit is the fitness value; W A is the prediction accuracy weight; A SVR is the support vector regression prediction accuracy; W F is the weight of the number of features; F k =1 means feature k is selected, F k =0 means feature k is not selected; K is the dimension of the initial feature set;

[0038] Perform crossover and mutation operations on chromosomes in the population;

[0039] Calculate the fitness value of each individual after mutation and the maximum fitness value in the population, and perform elite retention operations to obtain the next generation population;

[0040] The next generation population is used to continue evolution. When the evolutionary generation reaches the set maximum evolutionary generation, the evolution is stopped and the chromosomes of the current generation population are used as the parameter optimization values ​​of the input vector features, the support vector regression parameter C and the parameter γ that controls the kernel function scale.

[0041] Preferably, step S3 specifically includes:

[0042] S301, predicting the transformer oil temperature based on the transformer oil temperature prediction model, and calculating the prediction residual by the following formula:

[0043] e i =X i -X r,i

[0044] Where, e i is the i-th prediction residual; X i is the i-th predicted value; X r,i is the i-th actual value;

[0045] S302, calculate the transformer oil temperature meter's static identification identifier S based on the single-point deviation degree by the following formula: s1 for:

[0046] S s1 =IF{|e i |>10}

[0047] In the formula, IF is a logical judgment identifier. If |e i |>10, then S s1 =1, if |e i |≤10, then S s1 =0;

[0048] S303: Construct a residual sequence based on the prediction residuals, and calculate the root mean square error E of the prediction sequence using the following formula based on the residual sequence:

[0049]

[0050] Where n is the number of residual sequences;

[0051] S304, calculate the transformer oil temperature meter's meter state static identification identifier S based on the sequence fluctuation degree by the following formula: s2 for:

[0052] S s2 =IF{E>E limit}

[0053] Where, E limit is the preset root mean square error threshold. If E>Elimit, then S s2 =1, if E≤Elimit, then S s2 =0;

[0054] S305. Calculate the mean value e of the residual sequence by the following formula: avg and standard deviation σ e for:

[0055]

[0056]

[0057] S306, according to the average value e of the residual sequence avg and standard deviation σ e Calculate the skewness SK of the oil temperature residual series e , kurtosis KU e and standard deviation SD e They are:

[0058]

[0059]

[0060]

[0061] S307, according to the skewness SK of the oil temperature residual sequence e , kurtosis KU e and standard deviation SD e The Z values ​​of the skewness and kurtosis of the oil temperature residual series are calculated by the following formula:

[0062]

[0063]

[0064] Where, and are the Z values ​​of the skewness and kurtosis of the residual sequence respectively;

[0065] S308, comparing the Z value of the skewness of the residual sequence and the Z value of the kurtosis with the preset Z value threshold ranges of the skewness and the Z value of the kurtosis, respectively; if the Z value of the skewness of the residual sequence is within the preset Z value threshold range of the skewness or the Z value of the kurtosis is within the preset Z value threshold range of the kurtosis, executing step S309; ​​if the Z value of the skewness of the residual sequence is not within the preset Z value threshold range of the skewness and the Z value of the kurtosis is not within the preset Z value threshold range of the kurtosis, executing step S310;

[0066] S309, calculate the transformer oil temperature meter's static identification identifier S based on the residual statistical distribution of the meter state by the following formula: s3 , and go to step S311, where the formula is as follows:

[0067] S s3 =IF{|e i -e avg |>3σ e}

[0068] In the formula, if |e i -e avg |>3σ e , then S s3 =1, if |e i -e avg |≤3σ e , then S s3 =0;

[0069] S310, establish the probability density function Pr of the residual sequence, obtain the confidence interval We when the confidence level is 0.95, and calculate the static identification identifier S of the transformer oil temperature meter based on the residual statistical distribution by the following formula: s3 for:

[0070] S s3 =IF{e i >W2 or e i <W1}

[0071] Where W e ={[W1,W2]|Pr(W1≤e i ≤W2)=0.95}, where if e i >W2 or e i <W1, then S s3 =1, if W1≤e i ≤W2, then S s3 =0;

[0072] S311. Identify the state of the transformer oil temperature meter from the perspectives of single-point deviation, sequence fluctuation, and residual distribution based on the oil temperature prediction result, and obtain the static identification result of the transformer oil temperature meter state through the following formula:

[0073] S s =max(S s1 ,S s2 ,S s3 )

[0074] Where S s is the result of state static identification, max is the maximum value identifier, if S s =1, it is determined that the static identification result of the transformer oil temperature meter is abnormal.

[0075] Preferably, step S4 specifically includes:

[0076] S401, using the sliding time window to monitor the oil temperature prediction results, and obtain the oil temperature actual value sequence in the time window at time i as [X r,i-n+1 X r,i-n+2 … X r,i ], the oil temperature prediction value sequence is [X i-n+1 X i-n+2 … X i ], the oil temperature residual sequence is [e i-n+1 e i-n+2 … e i ];

[0077] S402, calculate the root mean square error E of the oil temperature prediction value sequence in the time window at time i by the following formula i , the mean value of the residual sequence e avg,i and standard deviation σ e,i for:

[0078]

[0079]

[0080]

[0081] S403, obtain the latest prediction residual e through the sliding window i , calculate the transformer oil temperature meter status dynamic identification identifier S by the following formula d1 for:

[0082] S d1 =IF{|e i |>10}

[0083] In the formula, if |e i |>10, then S d1=1, if |e i |≤10, then S d1 =0;

[0084] S404, based on the root mean square error E of the oil temperature prediction value sequence i The transformer oil temperature meter status dynamic identification identifier S is calculated by the following formula d2 for:

[0085] S d2 =IF{E i >E limit}

[0086] In the formula, if E i >E limit , then S d2 =1, if E i ≤E limit , then S d2 =0, where E limit =k E *E max , where k E is the threshold coefficient of the root mean square error, E max is the maximum value of the root mean square error;

[0087] S405: Obtain the latest prediction residual e through the sliding window i , the mean value of the residual sequence e avg,i and standard deviation σ e,i , and calculate the skewness SK of the oil temperature residual series e , kurtosis KU e and standard deviation SD e , the Z values ​​of the skewness and kurtosis of the oil temperature residual series are calculated by the following formula:

[0088]

[0089]

[0090] Where, and are the Z values ​​of the skewness and kurtosis of the residual sequence respectively;

[0091] S406: Compare the Z value of the skewness of the residual sequence and the Z value of the kurtosis with the preset Z value of the skewness and Z value of the kurtosis threshold ranges respectively. If the Z value of the skewness of the residual sequence is within the preset Z value threshold range of the skewness or the Z value of the kurtosis is within the preset Z value threshold range of the kurtosis, calculate the transformer oil temperature meter state dynamic identification identifier S by the following formula: d3 for:

[0092] S d3 =IF{|ei -e avg,i |>3σ e,i}

[0093] In the formula, if |e i -e avg,i |>3σ e,i , then S d3 =1, if |e i -e avg,i |≤3σ e,i , then S d3 =0;

[0094] If the Z value of the residual sequence skewness is not within the preset Z value threshold range of the skewness and the Z value of the kurtosis is not within the Z value threshold range of the kurtosis, the transformer oil temperature meter state dynamic identification identifier S is calculated by the following formula: d3 for:

[0095] S d3 =IF{e i >W2 or e i <W1}

[0096] Among them, if e i >W2 or e i <W1, then S d3 =1, if W1≤e i ≤W2, then S d3 =0;

[0097] S407, calculate the transformer oil temperature meter state dynamic identification identifier S by the following formula d4 for:

[0098] S d4 =IF{|e avg ,i|>e avg,limit}

[0099] In the formula, if |e avg,i |>e avg,limit , then S d4 =1, if |e avg,i |≤e avg,limit , then S d4 =0, where e avg,limit =k avg *|e avg,max |, where k avg is the threshold coefficient of the average value, e avg,max is the maximum value of the average value;

[0100] S408, the standard deviation σ of the residual sequence calculated by the sliding window e,i , calculate the transformer oil temperature meter status dynamic identification identifier S by the following formulad5 for:

[0101] S d5 =IF{|σ e,i |>σ e,limit}

[0102] In the formula, if |σ e,i |>σ e,limit , then S d5 =1, if |σ e,i |≤σ e,limit , then S d5 =0; where σ e,limit =k σ *|σ e,max |, where k σ is the threshold coefficient of standard deviation, σ e,max is the maximum value of the standard deviation;

[0103] S409, dynamically identify the identifier S according to the status of the transformer oil temperature meter d1 、S d2 、S d3 、S d4 and S d5 , the state dynamic identification result of the transformer oil temperature meter is identified by the following formula:

[0104] S d =max(S d1 ,S d2 ,S d3 ,S d4 ,S d5 )

[0105] In the formula, if S d =1, it is determined that the status dynamic identification result of the transformer oil temperature meter is abnormal.

[0106] Preferably, step S5 specifically includes:

[0107] According to the static identification result S of the transformer oil temperature meter s and abnormal dynamic identification results S d The transformer oil temperature meter status can be identified by the following formula:

[0108] S=max(S s ,S d )

[0109] Wherein, if S=1, it is determined that the transformer oil temperature meter is in an abnormal state; if S=0, it is determined that the transformer oil temperature meter is in a normal state.

[0110] In a second aspect, the present invention provides a transformer oil temperature meter abnormality identification system, comprising:

[0111] The data acquisition module is used to obtain the oil temperature data, load data and ambient temperature data of the transformer and construct an oil temperature prediction sample set;

[0112] A training module is used to train the oil temperature prediction sample set based on a support vector regression model that has been subjected to feature selection and parameter optimization using a genetic algorithm, so as to construct a transformer oil temperature prediction model;

[0113] A static identification module is used to predict the transformer oil temperature based on the transformer oil temperature prediction model, calculate the single point deviation, sequence fluctuation and residual distribution of the transformer oil temperature meter according to the oil temperature prediction result, and obtain the corresponding static identification result of the transformer oil temperature meter;

[0114] The dynamic identification module is used to monitor the oil temperature prediction results using a sliding time window to obtain the changing trend of the transformer oil temperature residual sequence and calculate the abnormal dynamic identification results of the transformer oil temperature meter based on the changing trend of the transformer oil temperature residual sequence;

[0115] The meter status identification module is used to identify the status of the transformer oil temperature meter by integrating the static status identification results and abnormal dynamic identification results of the transformer oil temperature meter.

[0116] It can be seen from the above technical solutions that the present invention has the following advantages:

[0117] The present invention constructs an oil temperature prediction sample set through the oil temperature data, load data and ambient temperature data of the transformer, trains the oil temperature prediction sample set based on a support vector regression model with parameters optimized by a genetic algorithm, constructs a transformer oil temperature prediction model, and performs transformer oil temperature prediction. The static state identification result of the transformer oil temperature meter is calculated from three aspects: single point deviation, sequence fluctuation and residual distribution of the transformer oil temperature meter. The sliding time window is also used to continuously monitor the changing trend of the oil temperature residual sequence, and the abnormal dynamic identification result of the transformer oil temperature meter is calculated based on this. The static state identification result and the abnormal dynamic identification result of the transformer oil temperature meter are integrated to identify the state of the transformer oil temperature meter, thereby improving the accuracy and efficiency of abnormal identification of the transformer oil temperature meter. BRIEF DESCRIPTION OF THE DRAWINGS

[0118] Figure 1 A flowchart of a method for identifying an abnormality of a transformer oil temperature meter provided by an embodiment of the present invention;

[0119] Figure 2 A schematic structural diagram of a transformer oil temperature meter abnormality identification system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0120] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0121] For easier understanding, see Figure 1 The present invention provides a method for identifying abnormality of a transformer oil temperature meter, comprising the following steps:

[0122] S1. Obtain the transformer's oil temperature data, load data, and ambient temperature data to construct an oil temperature prediction sample set.

[0123] Among them, the transformer oil temperature data, load data and ambient temperature data are obtained by dispatching the EMS system. The current load, historical load, current ambient temperature, historical ambient temperature and historical oil temperature are used as the input of oil temperature prediction, and the oil temperature at future moments is used as the output of oil temperature prediction to construct an oil temperature prediction sample set.

[0124] Considering the cumulative thermal effect of load and the superimposed thermal effect of base oil temperature and ambient temperature, the current load, historical load, current ambient temperature, historical ambient temperature and historical oil temperature are selected as the prediction inputs of the current oil temperature. Specifically, the input features of the oil temperature prediction model are initially set to include the current load, historical load in the previous 2 hours (including the load at 8 moments in the previous 15, 30, 45, 60, 75, 90, 105 and 120 minutes), the current ambient temperature, historical ambient temperature in the previous 2 hours (including the ambient temperature at 8 moments in the previous 15, 30, 45, 60, 75, 90, 105 and 120 minutes), and the historical oil temperature in the previous 2 hours (the oil temperature at 8 moments in the previous 15, 30, 45, 60, 75, 90, 105 and 120 minutes).

[0125] S2. Based on the support vector regression model with genetic algorithm for feature selection and parameter optimization, the oil temperature prediction sample set is trained to build a transformer oil temperature prediction model.

[0126] S3. Based on the transformer oil temperature prediction model, the transformer oil temperature is predicted. According to the oil temperature prediction results, the single-point deviation, sequence fluctuation and residual distribution of the transformer oil temperature meter are calculated to obtain the corresponding static state identification result of the transformer oil temperature meter.

[0127] S4. Use a sliding time window to monitor the oil temperature prediction results to obtain the changing trend of the transformer oil temperature residual sequence, and calculate the abnormal dynamic identification result of the transformer oil temperature meter based on the changing trend of the transformer oil temperature residual sequence.

[0128] S5. Identify the status of the transformer oil temperature meter by integrating the static status identification results and abnormal dynamic identification results of the transformer oil temperature meter.

[0129] The present invention provides a method for identifying abnormalities of a transformer oil temperature meter. An oil temperature prediction sample set is constructed through the oil temperature data, load data and ambient temperature data of the transformer. The oil temperature prediction sample set is trained based on a support vector regression model whose parameters are optimized by a genetic algorithm. A transformer oil temperature prediction model is constructed, and transformer oil temperature prediction is performed. The static state identification result of the transformer oil temperature meter is calculated from three aspects: single-point deviation, sequence fluctuation and residual distribution of the transformer oil temperature meter. A sliding time window is also used to continuously monitor the changing trend of the oil temperature residual sequence. Based on this, an abnormal dynamic identification result of the transformer oil temperature meter is calculated. The static state identification result and the abnormal dynamic identification result of the transformer oil temperature meter are integrated to identify the state of the transformer oil temperature meter, thereby improving the accuracy and efficiency of abnormal identification of the transformer oil temperature meter.

[0130] In a specific embodiment, step S2 specifically includes:

[0131] S201, taking the oil temperature prediction sample set as the training sample set (x j ,y j ), j = 1, 2, ..., J, x∈R N , y∈R, where x j is an N-dimensional input vector, y j is the output vector, R N is the feature space of the input vector, J is the number of training samples, and the regression estimation function constructed in the high-dimensional space is:

[0132]

[0133] Where: w is the weight coefficient; b is the threshold; is a nonlinear mapping function;

[0134] S202. Convert the regression estimation function into the objective function and inequality constraints of the quadratic programming problem:

[0135]

[0136] st|y j -(w T x j +b)|≤ε,j=1,2,…,J

[0137] Where T is the transposition symbol and ε is the slack variable;

[0138] S203, introduce slack variable ξ j , transform the inequality constraints of the quadratic programming problem into:

[0139] |y j -(w T x j +b)|≤ε+ξ j

[0140] S204. Introduce the penalty factor C and convert the objective function of the quadratic programming problem into:

[0141]

[0142] S205. For the linear inseparable problem, the radial basis kernel function is used to perform data mapping processing, and the result is:

[0143] K(x,x j )=exp(-γ||xx j || 2 )

[0144] Where γ is the parameter that controls the scale of the kernel function;

[0145] S206 , optimizing the input vector features, the support vector regression parameter C, and the parameter γ of the control kernel function scale based on a genetic algorithm, and performing oil temperature prediction based on the optimized transformer oil temperature prediction model.

[0146] The steps of optimizing the input vector features, the support vector regression parameter C, and the parameter γ for controlling the kernel function scale based on the genetic algorithm specifically include:

[0147] S2061. Randomly generate binary codes to form an initial population. The chromosomes in the initial population are composed of three parts: input vector features, support vector regression parameters C, and parameter γ that controls the kernel function scale.

[0148] S2062. Convert the binary code of the chromosome into a decimal value using the following formula:

[0149]

[0150] Where p is the decimal value of the parameter corresponding to the chromosome gene sequence; p min 、p max are the minimum and maximum values ​​of the parameters respectively; d is the decimal value of the chromosome gene sequence; l is the length of the chromosome gene sequence;

[0151] S2063. The fitness value of the chromosome is calculated by the fitness function of the following formula:

[0152]

[0153] Where, fit is the fitness value; W Ais the prediction accuracy weight; A SVR is the support vector regression prediction accuracy; W F is the weight of the number of features; F k =1 means feature k is selected, F k =0 means feature k is not selected; K is the dimension of the initial feature set;

[0154] S2064. Perform crossover and mutation operations on chromosomes in the population;

[0155] S2065. Calculate the fitness value of each individual after mutation and the maximum fitness value in the population, and perform elite retention operations to obtain the next generation population;

[0156] S2066. Continue evolution using the next generation population. When the evolutionary generation reaches the set maximum evolutionary generation, stop evolution and use the chromosomes of the current generation population as the parameter optimization values ​​of the input vector features, support vector regression parameters C and parameter γ for controlling the kernel function scale.

[0157] In a specific embodiment, step S3 specifically includes:

[0158] S301, predicting the transformer oil temperature based on the transformer oil temperature prediction model, and calculating the prediction residual using the following formula:

[0159] e i =X i -X r,i

[0160] Where, e i is the i-th prediction residual; X i is the i-th predicted value; X r,i is the i-th actual value;

[0161] S302, calculate the transformer oil temperature meter's static identification identifier S based on the single-point deviation degree by the following formula: s1 for:

[0162] S s1 =IF{|e i |>10}

[0163] In the formula, IF is a logical judgment identifier. If |e i |>10, then S s1 =1, if |e i |≤10, then S s1 =0;

[0164] S303: Construct a residual sequence based on the prediction residuals, and calculate the root mean square error E of the prediction sequence using the following formula:

[0165]

[0166] Where n is the number of residual sequences;

[0167] S304, calculate the transformer oil temperature meter's meter state static identification identifier S based on the sequence fluctuation degree by the following formula: s2 for:

[0168] S s2 =IF{E>E limit}

[0169] Where, E limit is the preset root mean square error threshold, if E>E limit , then S s2 =1, if E≤E limit , then S s2 =0;

[0170] S305. Calculate the mean value e of the residual sequence by the following formula: avg and standard deviation σ e for:

[0171]

[0172]

[0173] S306, according to the average value e of the residual sequence avg and standard deviation σ e Calculate the skewness SK of the oil temperature residual series e , kurtosis KU e and standard deviation SD e They are:

[0174]

[0175]

[0176]

[0177] S307, according to the skewness SK of the oil temperature residual sequence e , kurtosis KU e and standard deviation SD e The Z values ​​of the skewness and kurtosis of the oil temperature residual series are calculated by the following formula:

[0178]

[0179]

[0180] Where, and are the Z values ​​of the skewness and kurtosis of the residual sequence respectively;

[0181] S308, comparing the Z value of the skewness and the Z value of the kurtosis of the residual sequence with the preset Z value threshold ranges of the skewness and the Z value of the kurtosis, respectively. If the Z value of the skewness of the residual sequence is within the preset Z value threshold range of the skewness or the Z value of the kurtosis is within the preset Z value threshold range of the kurtosis, executing step S309; ​​if the Z value of the skewness of the residual sequence is not within the preset Z value threshold range of the skewness and the Z value of the kurtosis is not within the preset Z value threshold range of the kurtosis, executing step S310;

[0182] S309, calculate the transformer oil temperature meter's static identification identifier S based on the residual statistical distribution of the meter state by the following formula: s3 , and go to step S311, where the formula is as follows:

[0183] S s3 =IF{|e i -e avg |>3σ e}

[0184] In the formula, if |e i -e avg |>3σ e , then S s3 =1, if |e i -e avg |≤3σ e , then S s3 =0;

[0185] S310, establish the probability density function Pr of the residual sequence, obtain the confidence interval We when the confidence level is 0.95, and calculate the static identification identifier S of the transformer oil temperature meter based on the residual statistical distribution by the following formula: s3 for:

[0186] S s3 =IF{e i >W2 or e i <W1}

[0187] Where W e ={[W1,W2]|Pr(W1≤e i ≤W2)=0.95}, where if e i >W2 or e i <W1, then S s3 =1, if W1≤e i ≤W2, then S s3 =0;

[0188] S311. Identify the state of the transformer oil temperature meter from the perspectives of single-point deviation, sequence fluctuation, and residual distribution based on the oil temperature prediction result, and obtain the static identification result of the transformer oil temperature meter state through the following formula:

[0189] S s =max(S s1 ,S s2 ,S s3 )

[0190] Where S s is the result of state static identification, max is the maximum value identifier, if S s =1, it is determined that the static identification result of the transformer oil temperature meter is abnormal.

[0191] In a specific embodiment, step S4 specifically includes:

[0192] S401, using the sliding time window to monitor the oil temperature prediction results, and obtain the oil temperature actual value sequence in the time window at time i as [X r,i-n+1 X r,i-n+2 … X r,i ], the oil temperature prediction value sequence is [X i-n+1 X i-n+2 … X i ], the oil temperature residual sequence is [e i-n+1 e i-n+2 … e i ];

[0193] S402, calculate the root mean square error E of the oil temperature prediction value sequence in the time window at time i by the following formula i , the mean value of the residual sequence e avg,i and standard deviation σ e,i for:

[0194]

[0195]

[0196]

[0197] S403, obtain the latest prediction residual e through the sliding window i , calculate the transformer oil temperature meter status dynamic identification identifier S by the following formula d1 for:

[0198] S d1 =IF{|e i |>10}

[0199] In the formula, if |e i |>10, then S d1=1, if |e i |≤10, then S d1 =0;

[0200] S404, based on the root mean square error E of the oil temperature prediction value sequence i The transformer oil temperature meter status dynamic identification identifier S is calculated by the following formula d2 for:

[0201] S d2 =IF{E i >E limit}

[0202] In the formula, if E i >E limit , then S d2 =1, if E i ≤E limit , then S d2 =0, where E limit =k E *E max , where k E is the threshold coefficient of the root mean square error, E max is the maximum value of the root mean square error;

[0203] S405: Obtain the latest prediction residual e through the sliding window i , the mean value of the residual sequence e avg,i and standard deviation σ e,i , and calculate the skewness SK of the oil temperature residual series e , kurtosis KU e and standard deviation SD e , the Z values ​​of the skewness and kurtosis of the oil temperature residual series are calculated by the following formula:

[0204]

[0205]

[0206] Where, and are the Z values ​​of the skewness and kurtosis of the residual sequence respectively;

[0207] S406. Compare the Z value of the skewness and the Z value of the kurtosis of the residual sequence with the preset Z value of the skewness and the Z value of the kurtosis threshold range respectively. If the Z value of the skewness of the residual sequence is within the preset Z value threshold range of the skewness or the Z value of the kurtosis is within the preset Z value threshold range of the kurtosis, calculate the transformer oil temperature meter state dynamic identification identifier S by the following formula: d3 for:

[0208] S d3 =IF{|e i -eavg,i |>3σ e,i}

[0209] In the formula, if |e i -e avg,i |>3σ e,i , then S d3 =1, if |e i -e avg,i |≤3σ e,i , then S d3 =0;

[0210] If the Z value of the residual sequence skewness is not within the preset Z value threshold range of the skewness and the Z value of the kurtosis is not within the Z value threshold range of the kurtosis, the transformer oil temperature meter state dynamic identification identifier S is calculated by the following formula: d3 for:

[0211] S d3 =IF{e i >W2 or e i <W1}

[0212] Among them, if e i >W2 or e i <W1, then S d3 =1, if W1≤e i ≤W2, then S d3 =0;

[0213] S407, calculate the transformer oil temperature meter state dynamic identification identifier S by the following formula d4 for:

[0214] S d4 =IF{|e avg,i |>e avg,limit}

[0215] In the formula, if |e avg,i |>e avg,limit , then S d4 =1, if |e avg,i |≤e avg,limit , then S d4 =0, where e avg,limit =k avg *|e avg,max |, where k avg is the threshold coefficient of the average value, e avg,max is the maximum value of the average value;

[0216] S408, the standard deviation σ of the residual sequence calculated by the sliding window e,i , calculate the transformer oil temperature meter status dynamic identification identifier S by the following formula d5 for:

[0217] S d5 =IF{|σ e,i |>σ e,limit}

[0218] In the formula, if |σ e,i |>σ e,limit , then S d5 =1, if |σ e,i |≤σ e,limit , then S d5 =0; where σ e,limit =k σ *|σ e,max |, where k σ is the threshold coefficient of standard deviation, σ e,max is the maximum value of the standard deviation;

[0219] S409, dynamically identify the identifier S according to the status of the transformer oil temperature meter d1 、S d2 、S d3 、S d4 and S d5 , the state dynamic identification result of the transformer oil temperature meter is identified by the following formula:

[0220] S d =max(S d1 ,S d2 ,S d3 ,S d4 ,S d5 )

[0221] In the formula, if S d =1, it is determined that the status dynamic identification result of the transformer oil temperature meter is abnormal.

[0222] In a specific embodiment, step S5 specifically includes:

[0223] According to the static identification result S of the transformer oil temperature meter s and abnormal dynamic identification results S d The transformer oil temperature meter status can be identified by the following formula:

[0224] S=max(S s ,S d )

[0225] Wherein, if S=1, it is determined that the transformer oil temperature meter is in an abnormal state; if S=0, it is determined that the transformer oil temperature meter is in a normal state.

[0226] The above is a detailed description of an embodiment of a method for identifying an abnormality of a transformer oil temperature meter provided by the present invention. The following is a detailed description of an embodiment of a system for identifying an abnormality of a transformer oil temperature meter provided by the present invention.

[0227] For easier understanding, see Figure 2 The present invention provides a transformer oil temperature meter abnormality identification system, comprising:

[0228] The data acquisition module 100 is used to obtain the oil temperature data, load data and ambient temperature data of the transformer and construct an oil temperature prediction sample set;

[0229] A training module 200 is used to train an oil temperature prediction sample set based on a support vector regression model that has been subjected to feature selection and parameter optimization using a genetic algorithm, thereby constructing a transformer oil temperature prediction model;

[0230] The static identification module 300 is used to predict the transformer oil temperature based on the transformer oil temperature prediction model, calculate the single point deviation, sequence fluctuation and residual distribution of the transformer oil temperature meter according to the oil temperature prediction result, and obtain the corresponding static identification result of the transformer oil temperature meter;

[0231] The dynamic identification module 400 is used to monitor the oil temperature prediction results using a sliding time window to obtain the change trend of the transformer oil temperature residual sequence, and calculate the abnormal dynamic identification result of the transformer oil temperature meter based on the change trend of the transformer oil temperature residual sequence;

[0232] The meter status identification module 500 is used to identify the status of the transformer oil temperature meter by integrating the static status identification result and the abnormal dynamic identification result of the transformer oil temperature meter.

[0233] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0234] In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be through some interface, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0235] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0236] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0237] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for identifying abnormality of a transformer oil temperature meter, characterized in that: The following steps are involved: S1. Obtain transformer oil temperature data, load data, and ambient temperature data to construct an oil temperature prediction sample set; S2. Training the oil temperature prediction sample set based on a support vector regression model that has been subjected to feature selection and parameter optimization using a genetic algorithm to construct a transformer oil temperature prediction model; S3. Predicting the transformer oil temperature based on the transformer oil temperature prediction model, calculating the single-point deviation, sequence fluctuation, and residual distribution of the transformer oil temperature meter's residual sequence according to the oil temperature prediction result, and obtaining a corresponding static state identification result of the transformer oil temperature meter, including: S301, predicting the transformer oil temperature based on the transformer oil temperature prediction model, and calculating the prediction residual by the following formula: ； Where, e i is the i-th prediction residual; X i is the i-th predicted value; X r,i is the i-th actual value; S302, calculate the transformer oil temperature meter's static identification identifier S based on the single-point deviation degree by the following formula: s1 for: S s1 =IF{|e i |>10} In the formula, IF is a logical judgment identifier. If |e i |>10, then S s1 =1, if |e i |≤10, then S s1 =0; S303: Construct a residual sequence based on the prediction residuals, and calculate the root mean square error E of the prediction sequence using the following formula based on the residual sequence: ； Where n is the number of residual sequences; S304, calculate the transformer oil temperature meter's meter state static identification identifier S based on the sequence fluctuation degree by the following formula: s2 for: S s2 =IF{E>E limit } Where, E limit is the preset root mean square error threshold. If E>Elimit, then S s2 =1, if E≤Elimit, then S s2 =0; S305. Calculate the mean value e of the residual sequence by the following formula: avg and standard deviation σ e for: ； ； S306, according to the average value e of the residual sequence avg and standard deviation σ e Calculate the skewness SK of the oil temperature residual series e , kurtosis KU e and standard deviation SD e They are: ； ； ； S307, according to the skewness SK of the oil temperature residual sequence e , kurtosis KU e and standard deviation SD e The Z values ​​of the skewness and kurtosis of the oil temperature residual series are calculated by the following formula: ； ； Where, and are the Z values ​​of the skewness and kurtosis of the residual sequence respectively; S308, comparing the Z value of the skewness of the residual sequence and the Z value of the kurtosis with the preset Z value threshold ranges of the skewness and the Z value of the kurtosis, respectively; if the Z value of the skewness of the residual sequence is within the preset Z value threshold range of the skewness or the Z value of the kurtosis is within the preset Z value threshold range of the kurtosis, executing step S309; ​​if the Z value of the skewness of the residual sequence is not within the preset Z value threshold range of the skewness and the Z value of the kurtosis is not within the preset Z value threshold range of the kurtosis, executing step S310; S309, calculate the transformer oil temperature meter's static identification identifier S based on the residual statistical distribution of the meter state by the following formula: s3 , and go to step S311, where the formula is as follows: S s3 =IF{ } In the formula, if , then S s3 =1, if , then S s3 =0; S310, establish the probability density function Pr of the residual sequence, obtain the confidence interval We when the confidence level is 0.95, and calculate the static identification identifier S of the transformer oil temperature meter based on the residual statistical distribution by the following formula: s3 for: S s3 =IF{e i >W2 or e i <W1} Where W e ={[W1,W2]|Pr(W1≤e i ≤W2)=0.95}, where if e i >W2 or e i <W1, then S s3 =1, if W1≤e i ≤W2, then S s3 =0; S311. Identify the state of the transformer oil temperature meter from the perspectives of single-point deviation, sequence fluctuation, and residual distribution based on the oil temperature prediction result, and obtain the static identification result of the transformer oil temperature meter state through the following formula: S s =max(S s1 , S s2 , S s3 ) Where S s is the result of state static identification, max is the maximum value identifier, if S s =1, it is determined that the static identification result of the transformer oil temperature meter is abnormal; S4. Use a sliding time window to monitor the oil temperature prediction results to obtain the changing trend of the transformer oil temperature residual sequence. Calculate the abnormal dynamic identification results of the transformer oil temperature meter based on the changing trend of the transformer oil temperature residual sequence, including: S401, using the sliding time window to monitor the oil temperature prediction results, obtain The actual value sequence of oil temperature in the time window of time is , the oil temperature prediction value sequence is , the oil temperature residual series is ; S402, calculate by the following formula The root mean square error of the oil temperature prediction value series within the time window of time , the mean value of the residual series and standard deviation for: ； ； ； S403: Obtain the latest prediction residual through the sliding window , calculate the transformer oil temperature meter status dynamic identification identifier S by the following formula d1 for: S d1 =IF{| |>10} In the formula, if | |>10, then S d1 =1, if | |≤10, then S d1 =0; S404, based on the root mean square error of the oil temperature prediction value sequence The transformer oil temperature meter status dynamic identification identifier S is calculated by the following formula d2 for: S d2 =IF{ >E limit } In the formula, if >E limit , then S d2 =1, if ≤E limit , then S d2 =0, where E limit =k E *E max , where k E is the threshold coefficient of the root mean square error, E max is the maximum value of the root mean square error; S405: Obtain the latest prediction residual through the sliding window , the mean value of the residual series and standard deviation , and calculate the skewness of the oil temperature residual series , kurtosis and standard deviation , the Z values ​​of the skewness and kurtosis of the oil temperature residual series are calculated by the following formula: ； ； Where, and are the Z values ​​of the skewness and kurtosis of the residual sequence respectively; S406: Compare the Z value of the skewness of the residual sequence and the Z value of the kurtosis with the preset Z value of the skewness and Z value of the kurtosis threshold ranges respectively. If the Z value of the skewness of the residual sequence is within the preset Z value threshold range of the skewness or the Z value of the kurtosis is within the preset Z value threshold range of the kurtosis, calculate the transformer oil temperature meter state dynamic identification identifier S by the following formula: d3 for: S d3 =IF{ } In the formula, if , then S d3 =1, if , then S d3 =0; If the Z value of the residual sequence skewness is not within the preset Z value threshold range of the skewness and the Z value of the kurtosis is not within the Z value threshold range of the kurtosis, the transformer oil temperature meter state dynamic identification identifier S is calculated by the following formula: d3 for: S d3 =IF{ >W2 or <W1} Among them, if >W2 or <W1, then S d3 =1, if W1≤ ≤W2, then S d3 =0; S407, calculate the transformer oil temperature meter state dynamic identification identifier S by the following formula d4 for: S d4 =IF{| |>e avg,limit } In the formula, if | |>e avg,limit , then S d4 =1, if | |≤e avg,limit , then S d4 =0, where e avg,limit =k avg *|e avg,max |, where k avg is the threshold coefficient of the average value, e avg,max is the maximum value of the average value; S408, the standard deviation of the residual sequence calculated by the sliding window , calculate the transformer oil temperature meter status dynamic identification identifier S by the following formula d5 for: S d5 =IF{| |>s e,limit } In the formula, if | |>σ e,limit , then S d5 =1, if | |≤σ e,limit , then S d5 =0; where σ e,limit =k σ *|σ e,max |, where k σ is the threshold coefficient of standard deviation, σ e,max is the maximum value of the standard deviation; S409, dynamically identify the identifier S according to the status of the transformer oil temperature meter d1 、S d2 、S d3 、S d4 and S d5 , the state dynamic identification result of the transformer oil temperature meter is identified by the following formula: S d =max(S d1 , S d2 , S d3 , S d4 , S d5 ) In the formula, if S d =1, it is determined that the status dynamic identification result of the transformer oil temperature meter is abnormal; S5. Identify the status of the transformer oil temperature meter by integrating the static status identification results and abnormal dynamic identification results of the transformer oil temperature meter.

2. The method for identifying abnormality of a transformer oil temperature meter according to claim 1, characterized in that: Step S1 specifically includes: The transformer oil temperature data, load data and ambient temperature data are obtained through the dispatching EMS system. The current load, historical load, current ambient temperature, historical ambient temperature and historical oil temperature are used as the input of oil temperature prediction, and the oil temperature at future moments is used as the output of oil temperature prediction to construct an oil temperature prediction sample set.

3. The method for identifying abnormality of a transformer oil temperature meter according to claim 2, characterized in that: Step S2 specifically includes: S201, the oil temperature prediction sample set is used as a training sample set (x j ,y j ), j=1,2,…,J, , , where x j is an N-dimensional input vector, y j is the output vector, R N is the feature space of the input vector, J is the number of training samples, and the regression estimation function constructed in the high-dimensional space is: ； Where: w is the weight coefficient; b is the threshold; is a nonlinear mapping function; S202. Convert the regression estimation function into the objective function and inequality constraints of the quadratic programming problem: ； s.t. ; Where T is the transposition symbol, is the slack variable; S203, introduce slack variable ξ j , transform the inequality constraints of the quadratic programming problem into: ； S204. Introduce the penalty factor C and convert the objective function of the quadratic programming problem into: ； S205. For the linear inseparable problem, the radial basis kernel function is used to perform data mapping processing, and the result is: ； Where γ is the parameter that controls the scale of the kernel function; S206 , optimizing the input vector features, the support vector regression parameter C, and the parameter γ of the control kernel function scale based on a genetic algorithm, and performing oil temperature prediction based on the optimized transformer oil temperature prediction model.

4. The transformer oil temperature meter abnormality identification method according to claim 3 is characterized in that: The steps of optimizing the input vector features, the support vector regression parameter C, and the parameter γ controlling the kernel function scale based on the genetic algorithm include: Randomly generate binary codes to form the initial population. The chromosomes in the initial population are composed of three parts: input vector features, support vector regression parameters C, and parameter γ that controls the kernel function scale. Convert the chromosome from binary code to decimal value using the following formula: ； Where p is the decimal value of the parameter corresponding to the chromosome gene sequence; p min 、p max are the minimum and maximum values ​​of the parameters respectively; d is the decimal value of the chromosome gene sequence; l is the length of the chromosome gene sequence; The fitness value of the chromosome is calculated by the fitness function of the following formula: ； Where, fit is the fitness value; W A is the prediction accuracy weight; A SVR is the support vector regression prediction accuracy; W F is the weight of the number of features; F k =1 means feature k is selected, F k =0 means feature k is not selected; K is the dimension of the initial feature set; Perform crossover and mutation operations on chromosomes in the population; Calculate the fitness value of each individual after mutation and the maximum fitness value in the population, and perform elite retention operations to obtain the next generation population; The next generation population is used to continue evolution. When the evolutionary generation reaches the set maximum evolutionary generation, the evolution is stopped and the chromosomes of the current generation population are used as the parameter optimization values ​​of the input vector features, the support vector regression parameter C and the parameter γ that controls the kernel function scale.

5. The transformer oil temperature meter abnormality identification method according to claim 1, characterized in that: Step S5 specifically includes: According to the static identification result S of the transformer oil temperature meter s and abnormal dynamic identification results S d The transformer oil temperature meter status can be identified by the following formula: S=max(S s , S d ) Wherein, if S=1, it is determined that the transformer oil temperature meter is in an abnormal state; if S=0, it is determined that the transformer oil temperature meter is in a normal state.

6. A transformer oil temperature meter abnormality identification system, characterized in that: include: The data acquisition module is used to obtain the oil temperature data, load data and ambient temperature data of the transformer and construct an oil temperature prediction sample set; A training module is used to train the oil temperature prediction sample set based on a support vector regression model that has been subjected to feature selection and parameter optimization using a genetic algorithm, so as to construct a transformer oil temperature prediction model; A static identification module is used to predict the transformer oil temperature based on the transformer oil temperature prediction model, calculate the single point deviation, sequence fluctuation and residual distribution of the transformer oil temperature meter according to the oil temperature prediction result, and obtain the corresponding static identification result of the transformer oil temperature meter; The transformer oil temperature is predicted based on the transformer oil temperature prediction model. The single point deviation, sequence fluctuation and residual distribution of the transformer oil temperature meter are calculated according to the oil temperature prediction result to obtain the corresponding static state identification result of the transformer oil temperature meter, including: S301, predicting the transformer oil temperature based on the transformer oil temperature prediction model, and calculating the prediction residual by the following formula: ； Where, e i is the i-th prediction residual; X i is the i-th predicted value; X r,i is the i-th actual value; S302, calculate the transformer oil temperature meter's static identification identifier S based on the single-point deviation degree by the following formula: s1 for: S s1 =IF{|e i |>10} In the formula, IF is a logical judgment identifier. If |e i |>10, then S s1 =1, if |e i |≤10, then S s1 =0; S303: Construct a residual sequence based on the prediction residuals, and calculate the root mean square error E of the prediction sequence using the following formula based on the residual sequence: ； Where n is the number of residual sequences; S304, calculate the transformer oil temperature meter's meter state static identification identifier S based on the sequence fluctuation degree by the following formula: s2 for: S s2 =IF{E>E limit } Where, E limit is the preset root mean square error threshold. If E>Elimit, then S s2 =1, if E≤Elimit, then S s2 =0; S305. Calculate the mean value e of the residual sequence by the following formula: avg and standard deviation σ e for: ； ； S306, according to the average value e of the residual sequence avg and standard deviation σ e Calculate the skewness SK of the oil temperature residual series e , kurtosis KU e and standard deviation SD e They are: ； ； ； S307, according to the skewness SK of the oil temperature residual sequence e , kurtosis KU e and standard deviation SD e The Z values ​​of the skewness and kurtosis of the oil temperature residual series are calculated by the following formula: ； ； Where, and are the Z values ​​of the skewness and kurtosis of the residual sequence respectively; S308, comparing the Z value of the skewness of the residual sequence and the Z value of the kurtosis with the preset Z value threshold ranges of the skewness and the Z value of the kurtosis, respectively; if the Z value of the skewness of the residual sequence is within the preset Z value threshold range of the skewness or the Z value of the kurtosis is within the preset Z value threshold range of the kurtosis, executing step S309; ​​if the Z value of the skewness of the residual sequence is not within the preset Z value threshold range of the skewness and the Z value of the kurtosis is not within the preset Z value threshold range of the kurtosis, executing step S310; S309, calculate the transformer oil temperature meter's static identification identifier S based on the residual statistical distribution of the meter state by the following formula: s3 , and go to step S311, where the formula is as follows: S s3 =IF{ } In the formula, if , then S s3 =1, if , then S s3 =0; S310, establish the probability density function Pr of the residual sequence, obtain the confidence interval We when the confidence level is 0.95, and calculate the static identification identifier S of the transformer oil temperature meter based on the residual statistical distribution by the following formula: s3 for: S s3 =IF{e i >W2 or e i <W1} Where W e ={[W1,W2]|Pr(W1≤e i ≤W2)=0.95}, where if e i >W2 or e i <W1, then S s3 =1, if W1≤e i ≤W2, then S s3 =0; S311. Identify the state of the transformer oil temperature meter from the perspectives of single-point deviation, sequence fluctuation, and residual distribution based on the oil temperature prediction result, and obtain the static identification result of the transformer oil temperature meter state through the following formula: S s =max(S s1 , S s2 , S s3 ) Where S s is the result of state static identification, max is the maximum value identifier, if S s =1, it is determined that the static identification result of the transformer oil temperature meter is abnormal; The dynamic identification module is used to monitor the oil temperature prediction results using a sliding time window to obtain the changing trend of the transformer oil temperature residual sequence and calculate the abnormal dynamic identification results of the transformer oil temperature meter based on the changing trend of the transformer oil temperature residual sequence; The oil temperature prediction results are monitored using a sliding time window to obtain the changing trend of the transformer oil temperature residual sequence. Based on the changing trend of the transformer oil temperature residual sequence, the abnormal dynamic identification results of the transformer oil temperature meter are calculated, including: S401, using the sliding time window to monitor the oil temperature prediction results, obtain The actual value sequence of oil temperature in the time window of time is , the oil temperature prediction value sequence is , the oil temperature residual series is ; S402, calculate by the following formula The root mean square error of the oil temperature prediction value series within the time window of time , the mean value of the residual series and standard deviation for: ； ； ； S403: Obtain the latest prediction residual through the sliding window , calculate the transformer oil temperature meter status dynamic identification identifier S by the following formula d1 for: S d1 =IF{| |>10} In the formula, if | |>10, then S d1 =1, if | |≤10, then S d1 =0; S404, based on the root mean square error of the oil temperature prediction value sequence The transformer oil temperature meter status dynamic identification identifier S is calculated by the following formula d2 for: S d2 =IF{ >E limit } In the formula, if >E limit , then S d2 =1, if ≤E limit , then S d2 =0, where E limit =k E *E max , where k E is the threshold coefficient of the root mean square error, E max is the maximum value of the root mean square error; S405: Obtain the latest prediction residual through the sliding window , the mean value of the residual series and standard deviation , and calculate the skewness of the oil temperature residual series , kurtosis and standard deviation , the Z values ​​of the skewness and kurtosis of the oil temperature residual series are calculated by the following formula: ； ； Where, and are the Z values ​​of the skewness and kurtosis of the residual sequence respectively; S406: Compare the Z value of the skewness of the residual sequence and the Z value of the kurtosis with the preset Z value of the skewness and Z value of the kurtosis threshold ranges respectively. If the Z value of the skewness of the residual sequence is within the preset Z value threshold range of the skewness or the Z value of the kurtosis is within the preset Z value threshold range of the kurtosis, calculate the transformer oil temperature meter state dynamic identification identifier S by the following formula: d3 for: S d3 =IF{ } In the formula, if , then S d3 =1, if , then S d3 =0; If the Z value of the residual sequence skewness is not within the preset Z value threshold range of the skewness and the Z value of the kurtosis is not within the Z value threshold range of the kurtosis, the transformer oil temperature meter state dynamic identification identifier S is calculated by the following formula: d3 for: S d3 =IF{ >W2 or <W1} Among them, if >W2 or <W1, then S d3 =1, if W1≤ ≤W2, then S d3 =0; S407, calculate the transformer oil temperature meter state dynamic identification identifier S by the following formula d4 for: S d4 =IF{| |>e avg,limit } In the formula, if | |>e avg,limit , then S d4 =1, if | |≤e avg,limit , then S d4 =0, where e avg,limit =k avg *|e avg,max |, where k avg is the threshold coefficient of the average value, e avg,max is the maximum value of the average value; S408, the standard deviation of the residual sequence calculated by the sliding window , calculate the transformer oil temperature meter status dynamic identification identifier S by the following formula d5 for: S d5 =IF{| |>s e,limit } In the formula, if | |>σ e,limit , then S d5 =1, if | |≤σ e,limit , then S d5 =0; where σ e,limit =k σ *|σ e,max |, where k σ is the threshold coefficient of standard deviation, σ e,max is the maximum value of the standard deviation; S409, dynamically identify the identifier S according to the status of the transformer oil temperature meter d1 、S d2 、S d3 、S d4 and S d5 , the state dynamic identification result of the transformer oil temperature meter is identified by the following formula: S d =max(S d1 , S d2 , S d3 , S d4 , S d5 ) In the formula, if S d =1, it is determined that the status dynamic identification result of the transformer oil temperature meter is abnormal; The meter status identification module is used to identify the status of the transformer oil temperature meter by integrating the static status identification results and abnormal dynamic identification results of the transformer oil temperature meter.

Images